Trafiklab GTFS data

#---------------------------------------------------------------------------------------------------
# set up
#---------------------------------------------------------------------------------------------------

# clean
rm(list = ls())
invisible(gc())


options(dplyr.summarise.inform = FALSE)


# libraries
if (!require("pacman")) install.packages("pacman")
pacman::p_load(tidyverse, sf, sp, httr, mapview)



# avoid scientific notation
options(scipen=999)


# create directory
dir.create("data_input")
dir.create("data_output")
dir.create("output")

wd = getwd()
api = "Z:/"
data_input = paste0(wd,"/data_input")
data_output = paste0(wd,"/data_output")
output = paste0(wd,"/output")


# todays date, used as filter
today = str_remove_all(Sys.Date(), "-")


## Read API keys
api_fil <- read_file(paste0(api, "api"))

trafiklab_key = gsub('^.*trafiklab_gtfsstatik: \\s*|\\s*\r.*$', "", api_fil)

# url for GTFS
url <- paste0("https://opendata.samtrafiken.se/gtfs/ul/ul.zip?key=", trafiklab_key)

#---------------------------------------------------------------------------------------------------
# load data
#---------------------------------------------------------------------------------------------------

routes = read.csv2(paste0(data_input, "/trafiklab_ul/routes.txt"), 
                   sep = ",", encoding="UTF-8", stringsAsFactors=FALSE)

stops = read.csv2(paste0(data_input, "/trafiklab_ul/stops.txt"), 
                  sep = ",", encoding="UTF-8", stringsAsFactors=FALSE)

stop_times = read.csv2(paste0(data_input, "/trafiklab_ul/stop_times.txt"), 
                       sep = ",", encoding="UTF-8", stringsAsFactors=FALSE)

trips = read.csv2(paste0(data_input, "/trafiklab_ul/trips.txt"), 
                  sep = ",", encoding="UTF-8", stringsAsFactors=FALSE)

calendar_dates = read.csv2(paste0(data_input, "/trafiklab_ul/calendar_dates.txt"), 
                           sep = ",", encoding="UTF-8", stringsAsFactors=FALSE)



### Create filter variables

# service_id för rätt datum
service_id_inklud = calendar_dates %>% filter(date == today) %>% select(service_id) %>% pull()

# trips för rätt datum
trips_inklud = trips %>% filter(service_id %in% service_id_inklud) %>% select(trip_id) %>% pull()




#---------------------------------------------------------------------------------------------------
# Merge gtfs tables
#---------------------------------------------------------------------------------------------------

gtfs = stop_times %>%  
  left_join(., trips, by = "trip_id") %>%
  left_join(., stops, by = "stop_id") %>%
  left_join(., routes, by = "route_id") %>%
  mutate(hpl_id = substr(stop_id, 8, 13)) %>% 
  filter(trip_id %in% trips_inklud) %>%  # remove all rows referring to other dates
  distinct(arrival_time, departure_time, stop_id, .keep_all= TRUE) # remove duplicates

#---------------------------------------------------------------------------------------------------
# Data hantering
#---------------------------------------------------------------------------------------------------

antal_departure = gtfs %>% 
  group_by(hpl_id) %>% 
  summarise(antal_dep = n())

antal_linjer = gtfs %>% 
  distinct(hpl_id, route_short_name) %>% 
  group_by(hpl_id) %>% 
  summarise(antal_linjer = n())


## Tidtabelldata är på hållplatslägenivå. Ta medel för att skapa en koordinat per hållplats
hpl_koord = gtfs %>% 
  group_by(hpl_id, stop_name) %>% 
  summarise(lat = round(mean(as.numeric(stop_lat)), 5), lon = round(mean(as.numeric(stop_lon)), 5)) %>% 
  ungroup() %>% 
  left_join(antal_departure, by = "hpl_id") %>% 
  left_join(antal_linjer, by = "hpl_id") %>% 
  mutate(antal_dep_log = log(as.numeric(antal_dep)))

xy_gtfs = hpl_koord[,c("lon", "lat")]

spdf <- SpatialPointsDataFrame(coords = xy_gtfs, data = hpl_koord) # create spatial points

spdf1 = st_as_sf(spdf) %>% # convert to sf object
  st_set_crs(4326) 

Antal unika linjer per hållplats per vardagsdygn

mapview(spdf1, zcol = "antal_linjer")

Antal avgångar per hållplats per vardagsdygn

mapview(spdf1, zcol = "antal_dep_log")